The 1..T function is not working for me, and I don't know why. My code looks like this:
set TIME;
data;
set TIME = 1..8760 by 3;
display TIME;
Here I want it to display 1, 4, 7, 10, and so on, but it just goes 1..8760 by 3. How can I fix this so that I don't have to write 8,760 different numbers?
Thankful for answers!
I don't find any error in your code. I tried in my AMPL IDE:
AMPL IDE
I suppose that when you write "data" is that your code is in the .dat file and when you write "display" is your .run file
What are you using to write your model?? NEOS SERVER or the IDE??
Regards!
Try deleting the first two lines of your code.
It looks as if the format you're using to specify TIME only works in "model" mode; in "data" mode, the same text is interpreted as declaring a set of literals "3", "by", and "1..8760". (Quick test: if you type display card(TIME); you get a value of 3, telling you that there are exactly 3 members in this set.)
Section 5.2 of the AMPL Book recommends using a "x..y by z" type declaration in the model (with x, y, z as declared params) and then specifying values for x, y, z in the data.
Note that starting at 1 and increasing in steps of 3 won't actually get to 8760 exactly, so you might want to change your start to 0 or 3, or your end to 8761, if you want equal gaps between the numbers.
Related
I'm trying to finetune the pretrained Transformer-XL model transfo-xl-wt103 for a language modeling task. Therfore, I use the model class TransfoXLLMHeadModel.
To iterate over my dataset I use the LMOrderedIterator from the file tokenization_transfo_xl.py which yields a tensor with the data and its target for each batch (and the sequence length).
Let's assume the following data with batch_size = 1 and bptt = 8:
data = tensor([[1,2,3,4,5,6,7,8]])
target = tensor([[2,3,4,5,6,7,8,9]])
mems # from the previous output
My question is: I currently pass this data into the model like this:
output = model(input_ids=data, labels=target, mems=mems)
Is this correct?
I am wondering because the documentation says for the labels parameter:
labels (:obj:torch.LongTensor of shape :obj:(batch_size, sequence_length), optional, defaults to :obj:None):
Labels for language modeling.
Note that the labels are shifted inside the model, i.e. you can set lm_labels = input_ids
So what is it about the parameter lm_labels? I only see labels defined in the forward method.
And when the labels "are shifted" inside the model, does this mean I have to pass data twice (additionally instead of targets) because its shifted inside? But how does the model then know the next token to predict?
I also read through this bug and the fix in this pull request but I don't quite understand how to treat the model now (before vs. after fix)
Thanks in advance for some help!
Edit: Link to issue on Github
That does sound like a typo from another model's convention. You do have to pass data twice, once to input_ids and once to labels (in your case, [1, ... , 8] for both). The model will then attempt to predict [2, ... , 8] from [1, ... , 7]). I am not sure adding something at the beginning of the target tensor would work as that would probably cause size mismatches later down the line.
Passing twice is the default way to do this in transformers; before the aforementioned PR, TransfoXL did not shift labels internally and you had to shift the labels yourself. The PR changed it to be consistent with the library and the documentation, where you have to pass the same data twice.
I am new to SAS and have been using R most of the time. I am stuck with a simple and frustrating issue. All I want to do is to create a simple 3 X 3 matrix in SAS. But it throws an error. I need some help in understanding what's going on. The SAS documentation is not very helpful.
data matrixTest;
input Y $ X;
cards;
4 0
3 1
1 1
;
run;
/*Convert X to a categorical variable*/
data matrixTest;
set matrixTest;
if X = 0 then X = "0";
else X = "1";
run;
/*Get design matrix from the regression model*/
proc transreg data=matrixTest design;
model class(X/ zero=last);
output out=input_mcmc(drop=_: Int:);
run;
mX = {5 4 3, 4 0 4, 7 10 3};
And I get the following error when creating the matrix mX:
ERROR 180-322: Statement is not valid or it is used out of proper order.
Your error is that SAS is not a matrix language. SAS is more like a database language; the unit of operation is the dataset, analogous to a SQL table or a dataframe in R or Python.
SAS does have a matrix language built into the system, SAS/IML (interactive matrix language), but it's not part of base SAS and isn't really what you use in the context you're showing. The way you enter data as part of your program is how you did it in the first data step, with datalines.
Side note: You're also showing some R tendencies in the second data step; you cannot convert a variable's type that way. SAS has only 'numeric' and 'character', so you don't have 'categorical' data type anyway; just leave it as is.
Do not use the same data set name in the SET and DATA statements. This makes it hard to debug because you've destroyed your initial data set.
You cannot change types on the fly in SAS. If a variables i character it stays character.
If a variable is numeric, you assign values without quotes, quotes are used for character variables.
Your attempt to create a categorical variable doesn't make sense given the fact that it's already 0/1. Make sure your test data is reflective of your actual situation.
I'm not familiar with PROC TRANSREG so I cannot comment on that portion but those are the issues you're facing now.
As someone else mentioned, SAS is not a matrix language, it processes data line by line instead which means it can handle really, really large data sets because it doesn't have to load it into memory.
Your data set, matrixTest is essentially a data set and ready to go. You don't need to convert it to a matrix or 'initialize' it.
If you want a data set with those values then create that as a data set:
data mx;
input var1-var3;
cards;
5 4 3
4 0 4
7 10 3
;
run;
How can I have different parameters value defined in .ini file for each repeat in omnet using cmdenv? I have repeat value as 4 and trying to have different value of accidentStart and accidentDuration.
You can't. And shouldn't. The whole point of repetition is that all parameters have the same value, just the RNGs are seeded differently. So you get a different sample of the same distribution for every result value.
What you're looking for are iteration variables.
Something like this:
**.accidentStart = ${100, 200, 350}s
This will generate 3 runs without repetition, and 12 runs with repeat=4.
and if you add
**.accidentDuration = ${duration=300, 450, 600..1800 step 600}s
this will multiply the number of runs by another factor of 5.
By default, iteration variables produce a Cartesian product of their respective assigned sets of values. But there are ways to change this, consult the manual for how.
I am attempting to use fortran to write out a comma-delimited file for import into another commercial package. The issue is that I have an unknown number of data columns. My output needs to look like this:
a_string,a_float,a_different_float,float_array_elem1,float_array_elem2,...,float_array_elemn
which would result in something that might look like this:
L1080,546876.23,4325678.21,300.2,150.125,...,0.125
L1090,563245.1,2356345.21,27.1245,...,0.00983
I have three issues. One, I would prefer the elements to be tightly grouped (variable column width), two, I do not know how to define a variable number of array elements in the format statement, and three, the array elements can span a large range--maybe 12 orders of magnitude. The following code conceptually does what I want, but the variable 'n' and the lack of column-width definition throws an error (of course):
WRITE(50,900) linenames(ii),loc(ii,1:2),recon(ii,1:n)
900 FORMAT(A,',',F,',',F,n(',',F))
(I should note that n is fixed at run-time.) The write statement does what I want it to when I do WRITE(50,*), except that it's width-delimited.
I think this thread almost answered my question, but I got quite confused: SO. Right now I have a shell script with awk fixing the issue, but that solution is...inelegant. I could do some manipulation to make the output a string, and then just write it, but I would rather like to avoid that option if at all possible.
I'm doing this in Fortran 90 but I like to try to keep my code as backwards-compatible as possible.
the format close to what you want is f0.3, this will give no spaces and a fixed number of decimal places. I think if you want to also lop off trailing zeros you'll need to do a good bit of work.
The 'n' in your write statement can be larger than the number of data values, so one (old school) approach is to put a big number there, eg 100000. Modern fortran does have some syntax to specify indefinite repeat, i'm sure someone will offer that up.
----edit
the unlimited repeat is as you might guess an asterisk..and is evideltly "brand new" in f2008
In order to make sure that no space occurs between the entries in your line, you can write them separately in character variables and then print them out using theadjustl() function in fortran:
program csv
implicit none
integer, parameter :: dp = kind(1.0d0)
integer, parameter :: nn = 3
real(dp), parameter :: floatarray(nn) = [ -1.0_dp, -2.0_dp, -3.0_dp ]
integer :: ii
character(30) :: buffer(nn+2), myformat
! Create format string with appropriate number of fields.
write(myformat, "(A,I0,A)") "(A,", nn + 2, "(',',A))"
! You should execute the following lines in a loop for every line you want to output
write(buffer(1), "(F20.2)") 1.0_dp ! a_float
write(buffer(2), "(F20.2)") 2.0_dp ! a_different_float
do ii = 1, nn
write(buffer(2+ii), "(F20.3)") floatarray(ii)
end do
write(*, myformat) "a_string", (trim(adjustl(buffer(ii))), ii = 1, nn + 2)
end program csv
The demonstration above is only for one output line, but you can easily write a loop around the appropriate block to execute it for all your output lines. Also, you can choose different numerical format for the different entries, if you wish.
How does Mathematica decide when to round numbers in its output? For example,
giving the input
250000.5
gives the output
2500001
While
25000.5
is indeed printed as
25000.5
N[] isn't helpful here either, I need to use NumberForm[] to get it to actually print 250000.5 as 250000.5
I'm a Mathematica newbie, and I'm sure its ridiculously easy to control this threshold for when it starts ignoring decimals in its output, but could somebody please point me in the right direction?
another option for you to try, you can go to options and change the default PrintPrecision from 6 to say 16, and now you will see that it will print what you typed above
after I changed that to 16 (click on the field itself, and type 16 into the field to replace the 6, and hit return), then
Nasser is correct that PrintPrecision is the right setting.
You have a number of options for its use. You can set it Globally or for the specific Notebook using the Options Inspector. You can also use it directly with Style:
Style[250000.5, PrintPrecision -> 10]
250000.5
You can set it temporarily for one session like this:
SetOptions[$FrontEndSession, PrintPrecision -> 10]
Finally you can set it using Style Sheets (select cell type Output).
In the default TraditionalForm and StandardForm output modes Mathematica only shows a certain number of most significant digits. You can use InputForm to get the full precision number.